def test_thetas_size():
errors = []
desired_layers = [4, 3, 4, 3, 3]
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
for i in xrange(len(desired_layers) - 1):
if not nn._theta[i].size == desired_layers[i + 1]:
errors.append("error")
assert not errors
def test_activations_size():
errors = []
desired_layers = [4, 3, 4, 3, 3]
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
for i in xrange(len(desired_layers)):
if not nn._activations[i].size == desired_layers[i]:
errors.append("error")
assert not errors
def test_layers_sizes():
errors = []
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
if not nn._layers == [4, 3, 4, 3, 3]:
errors.append("error")
if not nn._C == 5:
errors.append("error")
assert not errors
def test_weights_size():
errors = []
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
if not len(nn._weights) == 4:
errors.append("error")
for i in xrange(1, 5):
if nn._weights[i - 1].shape != (nn._layers[i - 1], nn._layers[i]):
errors.append("error")
assert not errors
#! /usr/bin/python
# -*- coding: utf-8 -*-
import matplotlib.pyplot as plt
import aux_funcs as aux
import numpy as np
import classes as g_c
if __name__ == '__main__':
my_seed = 42 #SEED
max_iter = 300 #Max iterations for training
learning_rate = 0.02 #learning rate (alfa)
print "Start dummy test with XORG problem..."
nn = g_c.MLPerceptron(2, [2], 1)
print "MLPerceptron created with layers ", nn._layers
classes = np.array([0,0,1,1])
data = np.array([np.array([0, 0]), np.array([1, 1]), np.array([1, 0]), np.array([0, 1])])
print "Classes and data ready"
print "Starting training by hand..."
test = range(4)
for i in xrange(5000):
to_train = np.random.choice(test)
nn.train(classes[to_train], data[to_train], learning_rate)
print "trained"
def test_layers():
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
assert nn._layers == [4, 3, 4, 3, 3] and nn._C == 5
def test_net_input():
nn = cl.MLPerceptron(4, [3, 4, 3], 3)
nn.net_input(np.array([1, 2, 3, 4]))
assert np.array_equal(nn._activations[0], np.array([1, 2, 3, 4]))